Visual Masking

Visual Masking
The detection threshold (or the contrast corresponding to the threshold of perception) of a stimuli varies inversely as a function of its distance from the edge. This phenomenon is known as visual masking.

The implications of the visual masking for image coding are that large errors can be tolerated around edges (i.e., errors around edges can be quantized coarsely). In contrast, on a uniform background (smoothly varying regions or flat regions) small errors are detected. Therefore, if we wish to exploit the visual masking property of the human visual system the errors in the approximation should be confined around the edges.

Visual masking near an edge.